Aug 2026· Journal of Machine Learning Advances· Vol 1, pp. 1-25· 0 citations
TL;DR
FedGPA is proposed, a hierarchical framework in which clients with comparable resources and identical architectures form a group and a server mediates knowledge transfer across groups of diverse models, and at its core, the lightweight, model-agnostic Aligned Co-decision (Alco) Unit aligns class-level prototypical information across groups to bridge heterogeneous architectures.
Abstract
The heterogeneous nature has been regarded as a predominant challenge during the deployment of federated learning (FL) systems, wherein model heterogeneity—where clients train models of fundamentally different architectures—remains underexplored. Existing methods tolerate it poorly: they enforce interdependent model families, extract sub-models of one shared model, or rely on auxiliary public datasets and proxy models, which constrain model selection or demand storage that resource-limited clients cannot afford. We posit that an effective bridge across heterogeneous models should be both model-agnostic—decoupling knowledge transfer from incompatible parameter spaces—and storage-free, staying practical for clients with sharply different resources. On this basis we propose FedGPA, a hierarchical framework in which clients with comparable resources and identical architectures form a group and a server mediates knowledge transfer across groups of diverse models; at its core, the lightweight, model-agnostic Aligned Co-decision (Alco) Unit aligns class-level prototypical information across groups to bridge heterogeneous architectures, and a prototypical fusion step interpolates prototypes to regularize local training and refine cross-group knowledge. Experiments on CIFAR-10/100, EMNIST, and Tiny-ImageNet across heterogeneous architectures show that FedGPA consistently outperforms strong baselines from four method families, with the largest gains on resource-poor clients, at a cost of only ~100 KB of additional cross-group communication per round. A limitation of this study is that we provide no formal differential-privacy guarantee for the exchanged prototypes and do not address adversarial or data-quality attacks, leaving both to future work.
FedTopo is proposed, a relation-level framework that encodes global knowledge as class relation topology, capturing how classes relate within each client rather than where they lie in feature space.
FedSLM, a parameter-centric framework for federated fine-tuning with heterogeneous compressed clients, is proposed, which provides theoretical guarantees for adapter-level aggregation, subspace-alignment bounds for cross-group fusion, and a characterization of how the confidence loss mitigates weak-supervision noise.
Federated learning relies on aggregation schemes that assume all participants train models with identical architectures and a common parameter initialization. While this enables parameter-averaging strategies such as Federated Averaging, it also imposes a strong inductive bias by constraining local models to evolve from nearly identical starting points, potentially reducing model diversity and limiting exploration of the hypothesis space. Recent studies have explored shared-initialization-free and heterogeneous federated learning as largely independent research directions. In this context, we propose a shared-initialization-free, architecture-agnostic aggregation strategy based on meta-learning, where local models generate predictions over a reference dataset that are integrated to train a global meta-model. Among the different forms of heterogeneity in federated learning, this work focuses on statistical and model heterogeneity. The proposed framework is evaluated in both one-shot and multi-round federated settings against parameter-averaging and knowledge-distillation approaches while preserving the independent evolution of local models. Experiments across diverse federated settings, including homogeneous and heterogeneous models, varying numbers of participants, different levels of data imbalance on representative benchmark dataset demonstrate that the proposed strategy in a one-shot scenario achieves competitive performance under limited-data conditions, while iterative refinement of the prediction consensus in the multi-round setting yields consistent improvements over traditional aggregation strategies. These findings demonstrate that prediction-space meta-learning constitutes a practical alternative for federated aggregation without requiring shared parameter initialization. Future work will investigate more advanced consensus mechanisms, adaptive historical consensus strategies, and the evaluation of the proposed framework under larger-scale federated learning environments.
FedJigsaw is proposed, a novel framework that reshapes model personalization as a dynamic and decentralized model assembly problem that outperforms state-of-the-art MHFL baselines by up to 13.8% in relative accuracy while significantly shrinking cross-client performance variance, but also slashes decision-making latency and peak memory footprint compared to existing policy-driven methods.
Prototype-based knowledge sharing effectively mitigates data and model heterogeneity in federated learning (FL) by exchanging class-level semantic information. However, existing methods typically assume all local prototypes are equally reliable. Consequently, low-quality prototypes from heterogeneous models or dynamic clients can contaminate the global aggregation, leading to a vicious cycle of noise accumulation and performance degradation. To address this, we propose FedLEAF, a Federated Learning framework with server-side proactive Evaluation and clientside Adaptive Fusion. Specifically, the server employs an Adaptive Learning Prototypes (ALP) network to dynamically evaluate prototype reliability and generate learnable aggregation weights, ensuring that highquality prototypes exert a primary influence on the global model. Meanwhile, the client utilizes a Historical Consistency Fusion (HCF) strategy to selectively absorb global knowledge by assessing its consistency with locally maintained historical prototypes. Extensive experiments on standard datasets demonstrate that FedLEAF achieves effective improvements in model accuracy and robustness compared to existing methods.
Zhi-Yuan Zhu, Si-Yi Deng, Dapeng Wu et al.· 2026 International Conferenc...· 0 citations
